Continuous Data Mining to Sustain Advanced Process Control and Process Capability

Multiple factors can affect the performance of advanced controls.  These may be due to changes associated with the instrumentation, regulatory controls, or the process itself.  If not detected and diagnosed, the effect can be a partial loss of benefits, or in the worst case, the complete loss of benefits due to the controls being switched off.   It is therefore important to monitor not just control-specific metrics, but also process measurements and process metrics (KPIs) that may not directly be used by the advanced control system.  This presentation will describe a “big data small footprint” approach to statistically analyzing large amounts of historian data to determine when significant changes occur, and identifying candidate variables to facilitate cause and effect determination.  Plots and reports are automatically generated for control engineers, process engineers, and instrumentation technicians.   A benefit of the approach is bringing these groups together to solve problems and ensure process objectives are met.  Examples are presented to highlight the approaches and statistical techniques used.